Re: Short Vectors Versus Long Vectors

Thomas Koenig wrote:

MitchAlsup1 <mitchalsup@aol.com> schrieb:
 

Then contemplate for an instant that one would want SIMD instructions for
Complex numbers and Hamiltonian Quaterions......

 
Quaternions would be a bit over the top, I tink.  Complex
multiplication... implementing (e,f) = (a*c-b*d,a*d+b*c) is
 
         fmul    Rt1,Rc,Rb
         fmac    Re,Rd,Ra,Rt1
 
         fmul    Rt2,Rd,Rb
         fmac    Rf,Rc,Ra,-Rt2
 
So, you'd need both operands on both lanes.  Not very SIMD-friendly,
I would assume, but (probably) not impossible, either.
 

If you have the four operands spread across two SIMD registers, so
(Re,Im) in each, then you need an initial pair of permutes to make
flipped copies before you can start the fmul/fmac ops, right?
>
This is exactly the kind of code where Mitch's transparent vector
processing would be very nice to have.